Long noncoding RNA MEG3 inhibits oral squamous cell carcinoma progression via GATA3

Abstract

Oral squamous cell carcinoma (OSCC) accounts for about 90% of oral cancers. Expression of the long noncoding RNA (lncRNA) maternally expressed 3 (MEG3) has previously been reported to be downregulated in OSCC, and its overexpression can inhibit proliferation, migration, and invasion and promote apoptosis of OSCC cells. However, the mechanism underlying MEG3 downregulation in OSCC has not been well characterized. Here we report that low expression of MEG3 is caused by H3K27me3 modification of the MEG3 gene locus, and this is associated with the poor prognosis of OSCC. Overexpression of MEG3 inhibited the proliferation and invasion of OSCC cells. We observed that MEG3 was modified by m6A and bound to YTHDC1. Enhancer-controlled genes positively regulated by MEG3 were functionally enriched for the 'negative regulation of Wnt signaling pathway' term, as determined using metascape. GATA3 was predicted to be a transcription factor for these genes, and was demonstrated to bind to MEG3. Knockdown of GATA3 countered the effects on proliferation, invasion, and increased transcription of HIC1 and PRICKLE1 induced by MEG3 overexpression. In conclusion, our data suggest that MEG3 is downregulated in OSCC due to trimethylation of H3K27 at the MEG3 gene locus. The inhibitory effect of MEG3 on proliferation and invasion of OSCC cells was dependent on the binding of GATA3.

Keywords: MEG3; GATA3; H3K27me3; m6A; noncoding RNA; oral squamous cell carcinoma.

Fig. 1

Low expression of MEG3 was associated with poor prognosis in OSCC patients. (A) Analysis of MEG3 expression in 324 OSCC tissues and 32 normal control tissues based on TCGA data. Student's t‐test was applied for statistical analysis. Data are shown as median ± SD. *P < 0.05. (B) Analysis of the differences in MEG3 expression among G1, G2, and G3 tumor grades based on TCGA data. ANOVA followed by Tukey's test was performed for statistical analysis. Ns, nonsignificant. Solid line, median. Dotted line, quartile. (C) qRT‐PCR was performed to analyze MEG3 expression in 72 pairs of OSCC and adjacent normal tissues collected in this study. Student's t‐test was applied for statistical analysis. Data are shown as median ± SD. ****P < 0.0001. (D,E) overall survival (D) and progression‐free survival (E) of MEG3‐high (n = 36) and MEG3‐low (n = 36) groups based on the 72 patients enrolled in this study. The log‐rank Kaplan–Meier survival test was applied to compare the survival distribution of MEG3‐high and MEG3‐low groups.

Fig. 2

H3K27me3 modification of the MEG3 gene locus resulted in the downregulation of MEG3 in OSCC cells. (A) qRT‐PCR was used to measure the relative expression of MEG3 in SCC4, SCC9, SCC25, Cal27, and NOK cells. ANOVA followed by Tukey's test was performed for statistical analysis. *P < 0.05, **P < 0.01, vs. NOK cells. (B) Enrichment analysis of H3K27me3 signal of MEG3 gene locus in Cal27 and SCC4 cells based on the GSE149670 dataset. The H3K27me3‐enriched region was divided into region 1 and 2. (C,D) ChIP‐qPCR was used to detect the H3K27me3 modification of region 1 (C) and 2 (D) in Cal27 and SCC4 cells with or without GSK126 treatment. ANOVA followed by Tukey's test was applied for statistical analysis. **P < 0.01. (E) qRT‐PCR was used to measure the relative expression of MEG3 in Cal27 and SCC4 cells with or without GSK126 treatment. Student's t‐test was applied for statistical analysis. **P < 0.01. Experiments were performed in three biologically‐independent replicates. Data shown as mean ± SD.

Fig. 3

Overexpression of MEG3 inhibited the proliferation and invasion of OSCC cells. (A) qRT‐PCR was performed to detect the efficiency of MEG3 overexpression in Cal27 and SCC4 cells. ANOVA followed by Tukey's test was performed for statistical analysis. **P < 0.01. Ns, nonsignificant. (B,C) CCK8 assay was used to detect the proliferation of Cal27 (B) and SCC4 (C) cells in oe‐control and oe‐MEG3 groups. Student's t‐test was applied for statistical analysis. **P < 0.01. (D) Transwell assay was performed to detect the invasion of Cal27 and SCC4 cells in oe‐control and oe‐MEG3 groups. Student's t‐test was applied for statistical analysis. **P < 0.01. Experiments were performed in three biologically‐independent replicates. Data shown as mean ± SD. Scale bars = 200 μm.

Fig. 4

MEG3 was modified by m6A and bound to YTHDC1. (A) RIP‐qPCR was used to detect the binding between YTHDC1 and MEG3 in Cal27 and SCC4 cells. Student's t‐test was applied for statistical analysis. **P < 0.01. (B) Potential m6A modification sites of MEG3 were predicted using SRAMP (http://www.cuilab.cn/sramp). (C) RIP‐qPCR was used to detect MEG3 enriched levels of Cal27 and SCC4 cells in anti‐m6A and IgG groups. Student's t‐test was applied for statistical analysis. **P < 0.01. Experiments were performed in three biologically‐independent replicates. Data shown as mean ± SD.

Fig. 5

Screening and functional analysis of enhancer‐controlled genes positively regulated by MEG3. (A) Genes significantly correlated with MEG3 expression in OSCC tissues were screened based on TCGA data. (B) Intersections of genes positively correlated with MEG3 expression (blue, based on TCGA data) and enhancer‐controlled genes (yellow, based on GSE149670). (C) metascape (https://metascape.org/) analysis of the 576 intersecting genes. (D) H3K27ac signal at HIC1 and PRICKLE1 gene locus in Cal27 cells based on the GSE149670 dataset. (E,F) Pearson's correlation among MEG3, HIC1 (E) and PRICKLE1 (F) transcription in OSCC tissues was assessed based on TCGA data. (G,H) qRT‐PCR was performed to detect the relative transcriptional levels of HIC1 (G) and PRICKLE1 (H) in MEG3 overexpressed and control cells. Student's t‐test was applied for statistical analysis. **P < 0.01. Experiments were performed in three biologically‐independent replicates. Data shown as mean ± SD.

Fig. 6

The anticancer function of MEG3 in OSCC was dependent on the interaction with GATA3. (A) Transcription factors of the 576 intersecting genes were predicted using the toolkit for the cistrome data browser (http://dbtoolkit.cistrome.org/). (B) RIP‐qPCR was used to measure the binding between GATA3 and MEG3 in Cal27 and SCC4 cells. Student's t‐test was applied for statistical analysis. **P < 0.01. (C) Knockdown efficiency of GATA3 in Cal27 and SCC4 cells was assessed by qRT‐PCR. ANOVA followed by Tukey's test was performed for statistical analysis. **P < 0.01. Ns, nonsignificant. (D,E) qRT‐PCR was used to assess the relative transcription levels of HIC1 (D) and PRICKLE1 (E) in Cal27 and SCC4 cells after overexpression of MEG3 alone or overexpression of MEG3 with concomitant knockdown of GATA3. ANOVA followed by Tukey's test was performed for statistical analysis. **P < 0.01. (F,G) CCK8 assay was performed to detect the proliferation of Cal27 (F) and SCC4 (G) cells after overexpression of MEG3 alone or overexpression of MEG3 with concomitant knockdown of GATA3. ANOVA followed by Tukey's test was performed for statistical analysis. *P < 0.05, **P < 0.01. (H) Transwell assay was performed to detect invasion of Cal27 and SCC4 cells after overexpression of MEG3 alone or overexpression of MEG3 with concomitant knockdown of GATA3. ANOVA followed by Tukey's test was performed for statistical analysis. **P < 0.01. Experiments were performed in three biologically‐independent replicates. Data shown as mean ± SD. Scale bars = 200 μm.